Volume 9, Issue 1, Pages 16-23 (October 2014) De Novo Insertions and Deletions of Predominantly Paternal Origin Are Associated with Autism Spectrum Disorder  Shan Dong, Michael F. Walker, Nicholas J. Carriero, Michael DiCola, A. Jeremy Willsey, Adam Y. Ye, Zainulabedin Waqar, Luis E. Gonzalez, John D. Overton, Stephanie Frahm, John F. Keaney, Nicole A. Teran, Jeanselle Dea, Jeffrey D. Mandell, Vanessa Hus Bal, Catherine A. Sullivan, Nicholas M. DiLullo, Rehab O. Khalil, Jake Gockley, Zafer Yuksel, Sinem M. Sertel, A. Gulhan Ercan-Sencicek, Abha R. Gupta, Shrikant M. Mane, Michael Sheldon, Andrew I. Brooks, Kathryn Roeder, Bernie Devlin, Matthew W. State, Liping Wei, Stephan J. Sanders  Cell Reports  Volume 9, Issue 1, Pages 16-23 (October 2014) DOI: 10.1016/j.celrep.2014.08.068 Copyright © 2014 The Authors Terms and Conditions

Cell Reports 2014 9, 16-23DOI: (10.1016/j.celrep.2014.08.068) Copyright © 2014 The Authors Terms and Conditions

Figure 1 Experimental Overview (A) Indels were predicted in 787 families from the SSC using Dindel. Throughout the analytical pipeline, probands and siblings are treated equally to allow accurate assessment of de novo indel burden. Informative SNPs were used to establish the parent of origin of de novo indels. (B) Alignment errors at the end of reads lead to indels being miscalled as SNVs. (C) An indel can be represented in multiple ways in VCF files. See also Tables S1, S2, and S3. Cell Reports 2014 9, 16-23DOI: (10.1016/j.celrep.2014.08.068) Copyright © 2014 The Authors Terms and Conditions

Figure 2 De Novo Indel Burden and Genes with Multiple Hits (A) The rate of de novo indels and SNVs is shown for 602 probands (red) and matched unaffected siblings (blue). “All” refers to all RefSeq genes in hg19. “Brain” refers to the subset of genes that are brain expressed. “Nonsense” refers to single-nucleotide substitutions that result in a premature stop codon. “Splice site” refers to single-nucleotide substitutions that disrupt the canonical splice site. Error bars represent the 95% CIs and p values were calculated with a one-sided paired Wilcoxon test. (B) Two de novo frameshift indels in independent samples are shown in the gene KMT2E. Both indels are likely to induce nonsense-mediated decay (Nagy and Maquat, 1998). (C) Two de novo frameshift indels in independent samples are shown in the gene RIMS1. Both indels are likely to induce nonsense-mediated decay (Nagy and Maquat, 1998). See also Figures S1 and S2. Cell Reports 2014 9, 16-23DOI: (10.1016/j.celrep.2014.08.068) Copyright © 2014 The Authors Terms and Conditions

Figure 3 Sex Difference, Parent of Origin, and Parental Age (A) A consistently higher rate of de novo frameshift indels was observed in female probands (pink) compared with male probands (blue), but this difference was not observed in unaffected siblings. “All” describes all de novo frameshift indels and “Brain” indicates only those expressed in the brain. Error bars represent the 95% CIs and p values were calculated with a one-sided paired Wilcoxon test. (B) Histogram of full-scale IQ in all probands (green) and probands with a de novo frameshift indel (red). (C) The majority of de novo indels for which the parent of origin could be resolved were found to be on the paternal (blue) rather than the maternal (pink) chromosome (p < 0.001; binomial). This result was observed in both probands and siblings separately. (D) No clear relationship between the presence of a de novo indel and increased paternal age was observed for probands (green) or siblings (purple); p values were estimated with a Poisson regression. Cell Reports 2014 9, 16-23DOI: (10.1016/j.celrep.2014.08.068) Copyright © 2014 The Authors Terms and Conditions